“Nvidia, King’s College train robot overlords to spot oddities on radiology scans”
The actual headline that The Register UK used with its coverage of Nvidia’s AI partnership with King’s College London. AI marketers, this is what you’re up against.
Imaging Wire Sponsors
- Carestream – Focused on delivering innovation that is life changing – for patients, customers, employees, communities and other stakeholders.
- Focused Ultrasound Foundation – Accelerating the development and adoption of focused ultrasound.
- Medmo – Helping underinsured Americans save on medical scans by connecting them to imaging providers with unfilled schedule time.
- Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter.
- Pocus Systems – A new Point of Care Ultrasound startup, combining a team of POCUS veterans with next-generation technology to disrupt the industry.
- Qure.ai – Making healthcare more accessible by applying deep learning to radiology imaging.
The Imaging Wire
Philips and Medtronic’s Image-Guided Partnership
Philips took another step towards advancing its high-priority image-guided treatment business, signing an image-guided atrial fibrillation treatment collaboration deal with Medtronic. The alliance combines Philips’ KODEX-EPD cardiac imaging and navigation system (acquired through Philips’ 2018 EDP Solutions purchase) with Medtronic’s Arctic Front Advance cryoablation technology, while leveraging Medtronic’s Atrial Fibrillation Solutions division to take the solution to market.
KODEX-EPD helps guide electrophysiologists’ cryoablation procedures using dielectric imaging, which creates real-time CT-like 3D images of patients’ cardiac structures and reduces surgeons’ reliance on X-ray imaging. KODEX-EPD also confirms the positioning of Medtronic’s Achieve Mapping Catheter to help visualize pulmonary veins for pulmonary vein isolation (PVI) cryoablation procedures, while Philips is developing the KODEX-EPD’s balloon visualization capabilities to further expand its procedure support.
This seems like an ideal image-guided partnership, as it enhances Medtronic’s cryoablation capabilities and expands the reach of Philips’ KODEX-EPD system deeper within its target market.
ULRI-IE/IT, The Fluoroscopy Alternative
A Duke-led team shared compelling evidence in support of a new image modifier that allows a computer to show real-time movement of an instrument as it is adjusted, mimicking the use of fluoroscopy, but with far less radiation and much shorter procedure times. The image modifier, currently known as ultralow radiation imaging coupled with image enhancement and instrument tracking (ULRI-IE/IT), was studied in 23 randomized trials of nine procedures that require instrument localization, revealing that:
- Instrument localization time was 31.2% longer without ULRI-IE/IT assistance
- ULRI-IE/IT reduced the total number of images per case by 74.8%
- ULRI-IE/IT cut radiation exposure per case by 91.8%
- ULRI-IE/IT allowed physicians to place the instrument on their first attempt in 82.6% of trials, and 100% of second trials (vs. 4.65 images with unassisted fluoroscopy)
Although more research is needed, the team strongly supported ULRI-IE/IT for minimally invasive procedures as a way to reduce patient and surgeon radiation exposure and improve efficiency.
Overcoming Radiology AI’s “Last Mile” Challenge
A recent Nuance Healthcare Diagnostics Q&A detailed radiology AI’s “last mile” adoption challenge and outlined ways that Nuance and radiologists are overcoming these challenges.
- The Last Mile Challenge – Radiology AI’s “last mile” challenge has similarities with many efficiency breakthroughs that have emerged in different industries throughout history. Radiologists are realizing that AI tools can improve the quality and productivity of their work, but don’t believe they have the time to implement these tools.
- Last Mile Solutions – Nuance believes that the ability to seamlessly and intuitively integrate AI tools into current radiology workflows is vital to helping radiologists realize the benefits of AI. From a product design perspective, the key for this is to “think comprehensively,” such as delivering AI findings before a radiologist has read the study and dictated the report or making sure AI results are available alongside the images from the PACS and history from the patient’s EHR.
- AI Workflow Integration – Nuance sees simplifying AI development and deployment as key to workflow integration, specifically through a “unified market where developers can reach large numbers of radiology users who can easily discover and purchase new models,” like the Nuance AI Marketplace for Diagnostic Imaging. From there, access to the AI marketplace should be “integrated into the radiologist’s workflow tools, the worklist, the PACS and the Nuance PowerScribe reporting system” to support solution evaluation and then workflow integration.
- The Next Steps – There’s growing momentum toward the widespread adoption of AI in radiology workflows, but Nuance still sees challenges related to the AI reimbursement structure, access to diverse training data, and using previously-unavailable AI-generated data in radiology reporting.
Density Notification Laws Not Driving Breast Ultrasound Growth
Research from the University of Washington School of Medicine shared at ARRS and originally reported by Auntminnie.com found that the widespread adoption of breast density notification laws hasn’t resulted in a statistically significant increase in breast ultrasound usage.
The researchers found that between 2007 and 2015 ultrasound usage rates increased from 3.9% to 4.2% (1.1% relative increase) and US+DM rates increased from 7.1% to 10% (1.4% relative increase) in states that adopted breast density notification laws. These increases roughly align with previous research from Emory University that found follow-up breast ultrasounds after mammography screenings increased by 1.02% between 2007 and 2014 within states that adopted notification laws.
The UW study suggested that breast ultrasound adoption may increase at more substantial rates in the future. However, this growth may partially depend on how current and future notification laws are designed/redesigned, as the UW study found the most statistically significant breast ultrasound usage increases in states that require that women are specifically informed about their supplemental screening options.
inc42 on Qure.ai
A recent inc42.com profile on Qure.ai provided insights into the company, its growing portfolio of AI solutions, and market strategy.
- Creation and Leadership – Qure.ai was formed in 2016 by R&D leader Pooja Rao and CEO Prashant Warier with $30 million in funding from Fractal Analytics, where Warier also serves as Chief Data Scientist. Qure.ai further bolstered its leadership team in March, naming GE Healthcare veteran Chiranjiv Singh as its Chief Commercial Officer responsible for driving its global expansion.
- Solutions – Leveraging data from over 7 million images and backed by a number of validation studies, Qure.ai’s value proposition is based around diagnosis speed and accuracy. The company works with hospitals, health-centric non-profits, and medical devices companies with integrations in more than 12 countries (e.g. India, Philippines, US, France, Canada, Africa), targeting its qXR chest X-ray product at countries with high rates of TB and positioning its qER head CT solution for emergency departments in developed countries.
- Model – Qure.ai’s solutions are provided in a pay-per-use model, charging $1- $5 per scan depending on modality, allowing a wide range of healthcare providers adopt AI. Further supporting Qure.ai’s AI accessibility vision, its cloud-based solutions have minimal hardware requirements (e.g. $50 Raspberry Pi hardware kit) and are integrated with the 10 most-used radiology imaging platforms.
- Challenges – Like many in AI, Qure.ai sees overcoming market resistance and gaining customer and patient trust as keys to adoption. With that, Qure.ai places a major emphasis on ensuring that radiologists trust and see value in its solutions, including enabling interpretability by detailing “why and where the AI algorithm has detected an abnormality.”
The Wire
- United Imaging Healthcare announced the FDA clearance of the uPMR 790 HD TOF PET/MR scanner, highlighting its ability to scan a whole body within 20 minutes and perform simultaneous imaging with high-temporal and high isotropic spatial resolution. UIH leaned-in on the uPMR 790’s technology in this announcement, specifically its SiPM/LYSO-based PET platform (with 32cm PET FOV, 480 ps timing resolution, 2.8mm NEMA PET spatial resolution) that combines with its 3T MRI to allow “fast simultaneous whole-body PET and MR scans.” The system is intended for both clinical (staging, treatment planning, and response assessment) and research (theranostics and neuroscience) applications.
- A pair of closely-associated UK research teams published studies that suggest Whole-body MRI (WB-MRI) may perform better than the standard multi-modality imaging approaches for staging colorectal cancer and non-small-cell lung cancer (NSCLC). The two studies found that WB-MRI achieved sensitivity and final decision agreement rates that are close to multi-modality staging (slightly better for colorectal cancer, slightly worse for NSCLC), while reducing patient costs and staging time requirements in both studies.
- An American Medical Association survey reveals that, for the first time ever, more physicians work as employees of a hospital system or a practice owned by other physicians (47.4%) than run their own practice (45.9%). This is a big change since physician practice ownership peaked in 1983 (75.8%), but it’s not unexpected given the industry’s ongoing consolidation trend and the financial incentives that hospitals have to refer patients to hospital-owned practices. This employment trend is, of course, also taking place within radiology.
- The latest round of China tariffs kicked-in on Friday, increasing tariffs on $200 billion worth of Chinese imports from 10% to 25%. Although these tariffs largely targeted consumer products, they did include X-ray photographic plates and film and X-ray photographic film in rolls, joining a long list of medical imaging products and components that have been subject to 25% tariffs for nearly a year.
- NVIDIA and King’s College London announced a partnership to build an AI platform to improve the UK NIH’s radiology workflows. The initiative is part of KCL’s ongoing King’s London Medical Imaging & AI Centre for Value-Based Healthcare project and the latest announcement is directly connected to KCL’s fall 2018 adoption of NVIDIA’s DGX-2 AI research system and the NVIDIA Clara AI platform.
- The American College of Nurses-Midwives (ACNM) announced an alliance with GE Healthcare that will introduce its members to GE’s Vscan Extend handheld ultrasound through a series of educational webinars and training sessions. This is the latest example of GE’s training-focused approach to POCUS marketing, coming just three months after announcing a POCUS training collaboration with UltrasoundPodcast.com.
- One could see why the time-saving practice of using prepopulated default text in radiology reports may lead to biased reporting (radiologists erroneously using prepopulated impressions), a team from Cincinnati Children’s Hospital found no evidence of bias or any other negative impacts from using default text. The researchers looked at 248 ultrasound examinations for dysplasia of the hip (DDH – 472 hips) before adopting standardized reporting and studied 225 ultrasound examinations (426 hips) performed after the change, finding that a similar number of hips were diagnosed as DDH Graf 1 (74.4%, 75.4%) and Graf 2a (25.6%, 24.6%) before and after adopting standardized reporting.
- The FDA published a pair of X-ray guidance documents including a recommendation intended to make the U.S. X-ray regulation processes more consistent with the International Electrotechnical Commission (IEC) and a second guidance clarifying requirements related fluoroscopy irradiation time, last-image-hold, and emergency mode.
- Exact Imaging announced the CE Mark approval of its ExactVu micro-ultrasound system’s Abdominal Imaging Release, which includes its new EV5C Abdominal Transducer and Color Flow and Power Doppler capabilities. Exact Imaging suggests that the new solutions allow “urologists to perform complete urological ultrasound assessments including kidney examinations and biopsies, male pelvic and bladder examinations,” making it a “total men’s health imaging solution.”
- The British Columbian city of Kamloops may be great for winter sports, but it’s not ideal if you need an ultrasound, as a sonographer shortage (2.5 sonographers) has led to 10/12-month ultrasound wait times and forced some patients to drive 1.5-2.5 hours to neighboring cities for US scans. Kamloops may be a unique case, given its rural location, but sonographer shortages have been happening for some time, and considering ultrasound’s continued growth and diversification across departments/applications, it’s likely we’ll hear more talk of sonographer staffing issues going forward (it’s also a call for greater ultrasound efficiency).
The Resource Wire
–This is sponsored content.
- Carestream’s first OnSight 3D customer, Resurgens Orthopaedics (24 locations, 104 physicians), shared some of the benefits they’ve experienced from the cone beam CT system in this video, including the importance of weight bearing in surgery decisions, and the system’s image quality, ease of use, and fast study time.
- Did you know that radiologists’ follow-up recommendation rates can vary by between 2.8-fold and 6.8-fold depending on radiology division? Nuance’s PowerScribe Follow-up Manager and mPower Clinical Analytics solutions address follow-up inefficiencies by supporting departmental compliance efforts and helping ensure that the patients who received follow‑up recommendations actually return for their scans.
- The Focused Ultrasound Foundation’s 2018 Year in Review details the impressive research and clinical achievements that took place last year.
- We might be in the middle of Stroke Awareness Month, but not everyone is aware that strokes can cause the brain to lose over 30,000 cells for every second the brain goes without oxygen. Qure.ai’s qER triage aid detects critical problems in head CT scans in less than 10 seconds, helping doctors quickly prioritize the most urgent stroke and TBI cases.
- Did you know that imaging patients are most likely to no-show for their procedures on Mondays and Saturdays? By partnering with Medmo, imaging centers can keep their schedules full, despite the inevitable Monday no-shows.
- POCUS Systems is approved as a Veteran Owned Business with the US Government Office of Veterans Business Development, paving the way for partnerships with the federal healthcare delivery systems.
